It all depends on your relationship with your manager and your trust/feel/vibe for your employer.
If you have a positive relationship and trust 3-6 months is a fairly safe window that minimizes your risks. If you are unusual you might talk about it a year or more out (I've had these discussions especially in the context of performance and development). If you are in a role where you have an individualized employment contract you might also have specific language and terms there.
If you are unsure you just wait. The challenge is firing you after discussing retirement is potentially retaliation but fighting especially at retirement time is hard. Damages are limited as your "future wages" are small. Also watch out if you are covered by binding arbitration causes - arbitrators have shameful record of siding with employers way more than the courts and even when they find for the employee providing smaller awards when they do.
Consider financial factors on timing. On the delay side like any departure also watch out for big large period (quarterly/annual) payouts - bonuses, stock vesting, 401k match, etc. If in doubt wait until your next chunk due before retirement is in the bank. What is the financial health of the company? Are there layoffs coming? Could you potentially pick up a "package" on the way out? Is there a sabbatical or other long period benefit you want to receive before you go?
On the accelerate side do you have a lot of accrued PTO/vacation to be paid out that is potentially at risk? Do you normally carry a large "float" of expenses on personal cards/$$?
Consider talking to a lawyer (and tax planner if you haven't). You may have employment agreements that cover you into retirement that may impact you especially if you plan to work on the side. Also an employment law lawyer can advise you on exactly how to bring up the topic without giving "constructive notice" (effectively the employer constitutes your inquiry as an official notice of termination of employment).
In the end - be careful and put yourself first. Many a person has had last minute retirement plans change due to unexpected shifts in financial circumstance. In a healthy organization you should be able to have this conversation well in advance - but unfortunately the window of places this type of constructive 2-way dialog is possible is shrinking. On the other hand if your skill is critical and amenable to remote and part time work you might also work out a mutually beneficial arrangement for some post-retirement income as consulting. Do you have friends or family you might refer in later (not for bonus but just for the mutual benefit of the friend and family plus company)?
In the end you will have to make a choice on risk of early notice and associated early termination against the desire to do the right thing and keep a good relationship with your former employer and co-workers.
I'm a 10+ year medic myself in a Northern rural EMS system with an average response time of 10 minutes and an average transport time of 20-25 non-emergency and 15-20 emergency. That being said we had a large coverage area and were the last ALS service on the rural fringe of a large metro area so had outlying areas as well as mutual aid calls for neighboring BLS services. So 20+ minute response times were common and my personal longest 911 (not transfer) was 52 minutes. The transport time to a definitive care facility (versus community hospital) could be an hour in good weather, 90 minutes or more in bad.
The problem with these studies is they don't do a good job of factoring transport times in the analysis and typically are conducted in high density urban areas with response times under 10 minutes. The average urban transport time is 10-15 minutes...and some urban systems approach 5-8 for ALS for a large % of their population. In rural EMS 30-45 is uncommon and sub-10 minute response times are not common.
Another problem is urban environments with Fire based non-transport ALS where theres an additional transfer of ALS care on scene to the transport crew invariably burns time. In a well coordinated system that should be minimal and the fire based EMS often provides a 3-5 minute jump on scene arrival time for EMS (typically these systems then have a somewhat lower density of transport ALS units than a pure transport only service).
Keep in mind many of the European studies are done in countries with Doctors on the rigs where they often have scene times of an hour or more...that is clearly not productive. In the US even an ALS crew on a medical should have a scene time 20 minutes unless there's scene complications. That being said for a cardiac arrest patient unless there's a very close by ER there's very little value in transporting a patient prior to resuscitation unless the scene itself has hazards and difficulties (which happens plenty).
The relevant study also focuses on cardiac arrest only and ignores the impact on ALS on various medical issues short of full arrest like breathing, allergic and diabetic emergencies as well as non-arrest cardiac emergencies. An example is several studies including one I participated in showed good ALS with pre-hospital diagnosis of a STEMI with 12-lead ECGs lead to substantially reduced onset to intervention time with the cardiac surgical team (typically angioplasty) with corresponding increase in 12 month survival rates. Another aspect is changing the protocols and cultures to bypass small community hospitals for critical patients and take them straight to hospitals with definitive care (cath lab, trauma center, etc). There's a similar argument that once a patient is in good ALS care the value add of a non-definitive care facility rarely outweighs the 1-2 hour cost of the typical admit, eval, treat, transfer cycle.
One last problem is that the 12 month survival rate for pre-hospital arrest is not great in general and has only slowly improved so focusing on small differences versus the total picture does a disservice to the value provided by the system and individual providers.
If my wife/child/mother suffers a pre-hospital cardiac arrest I am quite sure I would much rather be in a place where the timely arrival of an ALS crew is part of the response.
I've worked for a variety of companies large and small and sometimes had to take a step sideways or backwards.
Rule #1 - The only real job security you have is your skills, network and ability to find work.
Example 1: I just switched jobs 3 months ago. I worked for a stable mid-market company in a 2nd tier job West Coast market with solid profitability, above average comp for the market for the top performers, good work/life balance, and a fairly flat revenue/market cap. Everybody there acted like they were lifers - the organization got 1/2 as much done as it could have, but everybody thought it was safe and stable because they made money hand over fist on their subscription based product and did just well enough to stay in the best of breed band in their market. But the reality is every stable, profitable company has someone willing to take it and squeeze it for more. That was nine months ago, 8 months ago they announced they were being taken private, 6 months ago the deal closed, and now it's official the company HQ is moving to a mid-America 3rd tier market and it's pretty clear that the timeline for the rest of the location is end of the year at best. And since the comp is above average in an above average market with new owners that have a financial model squarely centered on keeping labor costs (well) under 50th percentile they figure nobody will take the move and the paycut so pretty much no non-executives are being even offered a relocation. Everybody who was any good and had been there less than five years have already left - the rest are left wandering the beach aimlessly waiting for the tsunami to come to shore.
So...sometimes you have to move and take a "cut" in some ways. Just remember:
Rule #2 - Nothing not in writing maters. And even that is iffy.
Got bit by that once - and if they are are small or sleazy enough even if it's in writing it doesn't matter. No point in suing when 0 chance of recovering damages and they know it.
Example 2: Took a job to "build an embedded team" (in a Java shop) with a huge pay cut. Offer letter promised stock, six and twelve month salary adjustments, and at least 3 new hires within the first six months. Company was cool, team was great, office was great. We were downtown on the river, had a pub 40 feet down the hall. But there was no money. The next "round" never closed, the stock pool was exhausted at the time the letter was written, worked a year as a team of one and left. Would have bailed sooner except the job/team/office was very cool...loved coming to work every day...but was starving figuratively and somewhat literally.
So...weigh your options but long term it pays to shift and reinvent yourself once and a while. Myself and most of my peers recognize peoples ability to reinvent themselves on a resume...as long as it's not every year:-).
Reasons to make major changes in comp and/or size:
1. Move up in title. Many (most) people who make a senior management title or exec role do it in part by taking a shift down in size or salary to move up in title and then scale up from there.
2. Validate a title or role. Maybe you want to be a manager/director/architect. Maybe every company you have had a major role at is out of business or so small nobody has heard of them. Maybe you've lead a few teams here and there but didn't have that key word in the job title. You might go into a less interesting role to validate your resume at that level and title by taking the role at a larger or more recognizable company even if work is less interesting.
3. Make more money. Maybe you've done #1 and/or #2 and now you're finally being offered that director salary and role at a big company that throws money around. You're going into your peak earning years, your 401k has been ravaged by the past decade, go for it.
4. Have fun. Maybe you've done #3, or you were lucky with your investments or you picked the right startup. You've got retirement secured, your bored, go have fun.
I'm a pilot, paramedic and software engineer. My flying is personal but I try to take a "professional" approach. I agree that there are a lot of not so great pilots out there but the most basic pilot has had a bunch more training than 99.99% of the drivers out there. A typical "commercial" pilot with a commercial certification and an instrument rating (and typically multi-engine in both) has about 3-5 times the training typically required for a CDL.
Pilots and the flying industry are one of the most regulated endeavors in modern society. A good chunk of those regulations are "written in blood" from past accidents. Besides a few thousand pages of official FAA regulations there are thousdand more in ACs, TSOs and even industry standards like SAE, Milstd, ASTM, RTCC, etc. One big part of the problem is there are no standards for UAVs or UAV operators.
Flying is still heavily dependent on "see and avoid". The reality is we probably still avoid as many or more crashes from "big sky theory" than "see and avoid". The people who want to fly UAVs mostly want to fly them where the risk is highest - down low and over population areas. Also the UAV accident rate isn't as sparse as it sounds. There are well over 100,000 hours factoring in overseas usage. Even if you subtract out combat or unknown losses the accident rate of the UAV business is abysmal. Remember this is the industry that gave us unencrypted classified combat video. Check out http://www.homeland1.com/homeland-security-products/unmanned-aerial-vehicles-uav/articles/847069-accident-reports-show-us-drone-aircraft-plagued-with-problems/.
The argument that there is no pilot so the risk is minimal is disturbing. A predator is almost 30' long and a 48' wingspan - 1200# empty and over a ton fully loaded. This is comparable to most 4 seat trainers. Several of the private drones are smaller but have even less QA and little to no redundancy.
The first and only NSTB report on a drone crash is at http://www.ntsb.gov/ntsb/brief2.asp?ev_id=20060509X00531&ntsbno=CHI06MA121&akey=1. It cites the typical chain of errors as well as a series of poor design decisions. It also notes the wreckage path indicated a flat approach and a wreckage path of almost 100' with jet fuel scattered around the crash site (there was no ignition). The operations console being used had suffered approximately 16 lockups in the 4 months prior to the crash and suck lockups were viewed as normal and acceptable. The normal "lost link" procedure normally keeps the aircraft flying a predetermined route over unpopulated areas until control was restablished but improper recovery on the crash failure caused the engine to be turned off.
The lost link route procedure was called out in the NTSB report: "Another contractor, Organizational Strategies, Inc. (OSI), provided the coordinates for the lost-link waypoints to CBP. OSI reported that it developed the waypoints using an Internet satellite website. CBP reported that it also used the same Internet satellite website to verify the location of the waypoints. According to this website, some of the website's imagery is 1 to 3 years old. Neither OSI nor CBP used additional methods to confirm that the waypoints were not located over populated areas." No indication of the resolution of the satellite imagry used - and no requirement for direct verification.
In fairness the CBP is actually one of the more rigourous operators of UAVs. Their pilots are required to be certificated pilots with at least 200 hours of actualy flying time and 200 hours of UAS flying time. They also use specific TFRs to provide seperation and maintain contact and obtain clearances from ATC. Not all FAA "Certificates of Authority" require this level of coordination or training. Many smaller operators operate close enough to the ground or restricted terrain or existing restricted airspace viewed to not interfere with existing flight activity.
The simple reality is the UAV industry is about where manned flight was in the 30s. They hav
I'm going to avoid the topic of the relative goodness/badness of patents and address the question at hand.
Bonuses in my experience have been in the range of $1k-$2k for filing and $2k-$5k for granted. As noted above a big issues is what is the policy for patents granted to former employee inventors. I have missed out on all of my granted bonuses because all of my patents have been granted after I had moved on.
I would definitely have a cash component to the bounus - most of us have gotten savvy enough about the real odds on options payoff so it isn't necessarily a reward. At this point I view options as a quid pro quo for risk but not a substitute for fair salary and at least semi-reasonable workload.
The keeping in touch factor is a real issue - successful patents often involve one or more refilings and access to the inventor can be very helpful in adjusting claims or clarifying details. One former employer had the typical policy of not paying bounus for employees after they left. They had gotten back in touch with me about taking time to review a redone application which would have involved several hours of work. Given the lack of any reward compenent I requested a basic consulting rate for my time and they refused to I pretty much told them they were on their own.
A more recent former employer has a policy of paying bonuses to former employees if the employee can be reached within two years of the bonus event and providing they were cooperative in assisting any follow up reviews. Given that I have been cooperative and helpful with a couple of review and clarifications along the way.
Last but not least some places give a bonus premium on the "primary" versus additional inventors.
So in developing your policy pick amounts and make sure you are explicit either way about the bonus disposition for departed employees. The bonus progam can be an incentive for the innovators in the company as well as a tool to help get cooperation on the followup even if they have moved on. -- Mark
I'm in the same boat. I run a small shop with 10 folks and Exchange seems like overkill. Every year or so I go kicking around for an alternative and come back to staying put.
I'm not an MS bigot - there's FreeBSD and Linux (currently CentOS) doing the bulk of the server work here outside of groupware, but the key features in Exchange I have a hard time finding elsewhere is:
-The strong calendaring/scheduling including resource management
-The good quality web access (admittedly OWA requires IE to really shine)
-Support for mobile devices (between OMA/ActiveSync and IMAP I can almost always make it work)
-Easy to manage (at my small scale) replication
-Notes and Tasklists functionality does get used a lot in our organization
-Relatively seemless support for online/offline via Outlook
-Public folders w/support for offline and the various data items
-Rich, flexible and easy to manage ACLs on public and private folders
-We do a lot with shared calendars, notes and task lists
Various bits out there have pieces of the above but the real thing I can't seem to find is the strong calender/scheduling features with offline support. The reality is for serious road warriors the internet isn't anywhere close to 100% available. Remote areas, client sites with restrictive network access policies time in the airplane, all make 100% online solutions unworkable for us so far.
But as a small organization the only way we do this and stay legal is as a Microsoft partner through our NFR product licenses, the fact that we are 10 people makes it work. We couldn't afford the $1000s a year it would cost us to stay legal through retail or the lower volume license plans.
Like the rest of the world Exchange 2007 is on the shelf for now even though we have access to licenses for it. One is the 64 bit requirement along with overall performance loads raises the bar to deplying it. Microsoft is doing a typical MS thing and moving away from a good solution because of limitations before the next thing is ready. The de-emphasis of Public Folders and the threats to drop them completely are discouraging. Their answer is "SharePoint" which has no replication and no offline abilities currently - plus it forces me to a different tool and technology when all I need is to share access to various messaging items.
In practice I think MS is effectively priced itself out "small medium" businesses. The only way to deploy the Exchange, the required Windows server, CALs, etc. affordably is through a partner plan like us (which only scales to 10 users) or through Small Business Server - which doesn't scale effectively beyond 25-50 users and imposes deployment and licensing limitations that make it very difficult to move to the next level once you hit that ceiling.
Last but not least to that switch is data migration. I can kludge around on the messaging with various IMAP based solutions to moving the messages and but expect I'll have to write off the years of useful history in calendars. Also contact migration with LDIF still never seems to quite live up to promise, things always get munged, stripped or loose some supporting data.
There's plenty of slings and arrows at people who equate "Open" with "Free", and I agree you don't get something for nothing. Bur for many us it's a basic issue of affordability combined with legality. We work very hard to stay legal but even with our small shop licensing would eat us alive without the available partner program options and using open source in some areas. Everyone wants to grow their revenue and things creep up over time. Anti-spam, Anti-virus, office suites, server licenses, CALS, etc. Last time we ran numbers it would cost us over $2000/yr per/user to stay legal with retail licenses if we fully drank the MS cool aid. MS is getting pressure to prune their partner lists fr
Ooops, yeah, my bad, MM's correction is correct. But as AC notes, they have MySQL AB in a spot, especially now that they _also_ have bought SleepyCat/Berkeley DB.
If you read the attached article they're really talking about the performance deficit of PHP on Windows. Running PHP on Windows you have two primary server options - IIS and Apache, and then the various integration strategies (plug-in, CGI, FastCGI). In the case of IIS plug-ins with ISAPI there's a number of fiddling bits (like the ISAPI extension versus filter stuff) and getting all the settings right and quirky bits (example - I've never been able to mix PHP4 and PHP5 on one server with ISAPI). The most common setup is FastCGI - traditional CGI on Windows is painful because Windows process creation overhead is a lot more expensive - but FastCGI imposes tradeoffs on resources and performance and numbers of servers, etc.
All in all as a frequent user of PHP I figure this will benefit most of us. I worry a little less about extend and coopt here because I think Microsoft's primary motivation is to get you to stay on Windows Server/IIS even for your PHP platforms and then continue to try to push.NET down your throat.
As for MySQL - it's now owned by Oracle and IMHO Larry Ellison has a far better shot at being the antichrist than Bill Gates. Yes we have all that GPLd code but the company, talent and non-GPL rights to the code are owned by Oracle.
Another aside is both cases are examples that successful technology attracts attention and money regardless of the ideology or goals that spawned it. As someone who's been plugging away in development almost 20 years now I'd say the "success" part is tied more to market success and less to technological superiority - MySQL and PHP both end up strong on both counts, but neither would have attracted a whit's bit of interest if their market penetration was 2%.
I know this reply is way late in the arc on this post but the threads here hit home so very, very much I figured I'd post to an empty room anyway.
I've been on both sides of the fence as top talent and trying to hire top talent. First the hiring view. The reality is in certain niches (especially networking, protocol, embedded systems type talent) there aren't nearly as many rock stars as you think. My last permanent job was smack in the middle of the Valley and the reality is there were probably only a couple of hundred people out there in top little piece of the bell curve in any general skill set. There's a ton of talented people just a bit farther down the curve, but you look at most tech companies and it's really a very, very small group of people that drive the design, technology and innovation. It also seemed like you had to find the rare combination of technically brillant but market aware individual as the quality of product marketing in most organizations is _sh*t_ as far as defining innovation (versus spitting out glossies extolling your virtues and trashing the competition).
Then you can't hire them because you can't really get management to pay the top salaries because the VCs are breathing down the necks of the officers and board wondering why it hasn't already all been shipped to India much less trying to shake out an attractive salary.
On the working side once you get hired you end up only being around 2-4 years on average because you get squeezed on the late side of the capital/startup curve and salaries stagnate at 2-3% a year and bonuses and perks get squeezed - or you just plain run out of $$ and go under and if you're good enough you eventually become the most expensive body in the shop and easy pickings in a quarter the beancounters are sweating dollars instead of head count.
I've worked for a ton of startups as well as established companies and the reality is 98% of companies more than 10-20 people as an organization don't give a rat's a** about their people. There's still individuals out there that fight the good fight but the reality is if they need to do a short pump and dump on the share price or fixup the burn burn rate prior to the next financing round they'll RIF us 20% accross the board in a heartbeat and then whine a month later about the lack of employee loyalty these days and wonder why we jump ship for a lousy 5% salary differential or an extra week of vacation. Options are worthless - I've had some come in and most go under and my net gain from options over a 20 year career works out to about $1500 a year (maybe $2000 in today's dollars).
There's a glass ceiling of $100-$120k a year everywhere outside the Valley and maybe $140k a year in the Valley (where the delta doesn't cover the cost of living diffs) if you want to be a developer, architect, technical anything. And that's the very peak of the bell curve. There's a bunch of development jobs that have typical _Senior_ developer salaries down more in the $60-$80k and the non-developer IT guys have it worse. To move forward you have to cross the dark side to management or leave the traditional work force entirely and do your own thing - consulting, starting a business, inventing the next pet rock. Out here in middle America the number development focused technical companies is small enough that there aren't a lot of Director and up level positions - and the majority of them are in larger organizations that tend to promote from within so to break that inital barrier into management requires backsliding your salary - sometimes a lot - and dredging along at "Director of IT" for a salary that works out to a pitiful amount per hour just to slide the resume over to a new track with job titles with words like Director and Vice in them.
There's a lot of top flight technical talent outside the Valley and the Beltway that will choose to schluck along under-employed just to live in a place with affordable housing and decent public schools and a bit of green grass, but most companies are so weakl
I used to live in Pleasanton, California and would fly out of Livermore into the former Castle Airforce Base in Atwater, CA which was home to 93rd Bombardment Wing's B-52s for almost 30 years (Castle Airport).
It's now officially 11,802 by 150, in reality the runway is 13,000 by 300 but they striped it down to reduce maintenance costs.
It's an uncontrolled field but it does get some use for flight testing and training and the FBO there does do heavy maintenance so I have shared the pattern with a 737 once and a DC-10 another time.
The big empty B-52 hangers are open to walk through - it's interesting looking at the "wash stand" for B-52s.
When the Air Force was still there (up through 1997) they used to host the Golden West EAA Fly-Ins - for that they would split the runway in two and use each end at the same time.
There's a few 100 acres of paved tarmac and they do routinely close off part of it for road rallys and other driving schools.
There's a nice museum on field (Castle Air Museum) and the FBO will shuttle you there for free, the walk is about 1/2 mile. The museum itself runs a few bucks but is worth it.
I work part time as a paramedic. I agree with most posters, after 50-60 hours a week of startup grind the last thing I want to do is IT or programming for someone else. What little spare energy I have goes into my personal bits for myself.
Being a paramedic is everything IT is not - lot's of people interaction, some physical labor, outside time. It's also personally rewarding in a way that programming isn't. I've never been applauded for my work as a programmer, something I have had happen a few times as a medic. I've also never been offered a blow job as a programmer, but I have been a few times as a medic. As a married and ethical man I do not require applause nor accept the blow jobs but as a human being and a terminal male I appreciate the consideration present in both.
I originally became a medic because a bit of volunteering as an EMT showed me I loved the work and I thought that (as my plans were at the time) that the medic job would provide a salary baseline and benefits when consulting wore thin. Well - the bubble burst and I'm an employee again but I keep up the medic because it's a perfect escape from the office and if things really go south it's one job I know they _can't_ ship to India.
The wages are not great but when you can work a 24 hour shift and get paid 24 hours for one calendar day it does add up even for a few days a month enough for even an overpaid technoweenie like me to notice.
After 15 years in the development trenches I would love to work full time as a medic and have the energy to expend my skills part time on programming but then I couldn't afford a new GeForce 6600 or flying as often. I also like not sweating that I'm getting the $4 latte instead of the $1.25 cup of joe, something most full time medics have to worry about.
At some point I may make the trade of money versus time for lifestyle as we achieve certain financial goals, but for now it works as is.
I'm a private pilot and have several friends and relatives who work as pilots from the lowly cargo dogs flying single engine FedEx to a DC-10 captain at American Airlines.
One - the $250k number probably represents the top 5% of pilots for compensation. Even the pilots earning those numbers took 10-15 years and spent a whole lot of $$ to get there - makes Med School look fast and cheap.
Two - many (not all) pilots are paid all or mostly by the hour flown - they aren't paid for the hours of flight planning and prep or the days away from home. An international flight crew might easily be away from home 20 out of 30 days and that is gone for 24 hours, no family, maybe one phone call a day if the time you are on the ground lines up with the time your family is awake. Out of that remaining 10 days they may still have to get in their periodic training, reviews, etc.
Three - yes computers can land the airplane all the way to the pavement. But they can only do it (today) at a handful of runways at a handful of airports with the right precision (Cat III ILS) approach. As a point of reference in all of California - the home of the most airports overall and the most airports which scheduled air service of any of the US states there are less than a dozen Cat III ILS approaches. Even that number is unusually large because of the relatively small (towns like Fresno) cities with Cat III because of the unique weather of the Central Valley (Tule Fog).
Over the other 98% of the world you need a pilot to keep flying and land. Technology is rapidly improving and things like GPS will make more and more precision approaches possible but it will still take a pilot to land at the majority of runway ends in the world. And when shit goes wrong you want to land at the closest runway that you can get safely stopped in - not the one that the computer can use 30 minutes or an hour or more farther away.
Computers are great at doing things requiring fast calculations and precise responses but the/. crowd should know more than most how hard it is to program reponses for when the situation departs "normal" parameters.
Granted - people don't always do better - they typically still need some training and guidance for how to respond and tend to do poorly when flight departs completely from the norm. Look at the end result of events like the accidental thrust reverser deployments in 767s or the rudder failures in 737s. But in many of these events even the post-mortem reviews shows that the "window" to recover from these events is often 10 seconds or less. But other situations like the Souix City crash or the volcanic ash incident of Speedbird 1 (British Airways) show that the adaptability of a human crew is irreplacable.
It is probably inevitable that computers will do more and more of the flying - they already do 80%-90% of the enroute flying and they typically fly most of the hard appraochees (at least on the "big" guys) with the pilot acting more as a director and flight computer programmer than a pilot. But think - as the pilots really fly less and less then how do they maintain their skills for the 5% of the time that they are really needed?
So I think the pilots of the world earn every nickel the hard way.
In past lives I've worked with career spooks and even had a few of my own visits to the NSA (no I've never been to Fort Meade - any real spook groupie knows that the Infosec guys in "X" were up at the "Friendship Annex" by BWI...).
But it was an infrequent but not common practice at vendors that some systems to certain customers would be delivered by leaving a trailer at location X where it would disappear for a few days then they would get a call to pick up the trailer at Y.
There is also a verification process to sell to private and foreign customers to verify that systems aren't going to the wrong people.
But the real question is that now that Cray is Cray again will they start shipping a case of Leine's with every system like they used to? (Probably not given that manufacturing is no longer in Chippewa Falls, WI).
My take on the point of the original question was why/would RAM drives replace traditional disks. I'm basically arguing not on that point.
I'll also be the first to agree that in the cheap PC hardware era many performance problems can be most effectively solved by throwing hardware at them. An example would be a CPU bound application that needs to be twice as fast that runs on 1 PC server with 9 month old technology could easily make that target with a $3000-$5000 hardware upgrade. That doesn't by very many man hours to be sure.
I wasn't (and wouldn't) claim that a RAM disk wouldn't be faster than a platter based drive - only that the amount of that gain is a complex question. RAM drives for certain point applications (lock dirs, temporary files, caches, etc) are great. Even Platypus which you mention makes a RAM drive product that doesn't have persitent storage (and admittedly some that do).
So just a quick followup. I wouldn't have bothered except that it's my first post to be replied to and my first post to be moderated. Woo hoo.
Wireless is one - there are a number of 10BaseT to DS3 (ie: 45Mbps) products available on the Market. Link price runs $30,000-$60,000 (link=2X antennas+boxes, etc). These solutions generally top out at 10-20 miles, a few do 30. So you could do it in 2-3 hops. Look for products that operate in unlicensed spectrum (typically 2.4GHz and 5.8 GHz Uniband) to speed deployment. Harris Microwave (http://www.microwave.harris.com/) and Western Multiplex (http://www.wmux.com/) are two that come to mind.
Another option is free space optical (fso) - companies like Lightwave (www.lightwave.com).
Higher bandwidth wireless and require line of sight. RF in fact requires more clearance - based on a function called the Fresnel zone radius - a shape resembling a stretched football - that makes required altitudes for longer links even higher. General rule is that 60% of the 1st fresnel zone must be unobstructed. At 45 miles and 2.4 GHz the middle of the zone typically requires a clearance of 100'-160'. At 5.8GHz this drops to 60'-100' roughly (the function is proportional to wavelength and thus inversely proportional to frequency) - one calculator is at http://wireless.ydi.com/calc.asp#FresnelZone - another simpler one is at http://www.ingenious-nets.com/fresnel-calc.html.
But a number of people claim getting 20 mile ranges out of 802.11b at 2Mbps rates. One sample calculation gives the required antenna height at abou 175 feet (about 100 feet for line of sight and 75 feet for the fresnel zone clearance) (see http://www.oreillynet.com/cs/weblog/view/wlg/197).
A 175' antenna height is potentially very expensive. More expensive than typical antennas for normal RF applications because of the fact that the antenna itself must be very solid in wind to maintain line of sight alignment. But a some water towers or a 15 story building would do it.
Net is that for RF wireless you probably need to drop to more intermediate links to manage LOS and clearance.
But there may be viable wireline alternatives. One source of help would be a 'national' ISP like MCI/Worldcom (UUNet), Sprint/Earthlink, or Verizon (GTE) (in the case of the US). Their local sales forces are often very in touch with various alternative data carriers to provide the backhaul. Pipeline companies, dedicated fibre companies and power companies are among the entities that often have a datalink business.
Almost any local carrier can provide at least a T1 level frame relay or ATM circuit. One nice technology for this application is Inverse-Multiplexed ATM. For instance - you take 3 T1 ATM circuits and run IMA over them and the link looks like one ATM link at 4.5Mbps. If individual links go up or down the IMA link stays up - the bandwidth just goes up or down.
There are federal programs available to help with school internet access. Many states also have programs geared towards both education and general rural data access.
The advice about a Co-op is good. If you don't want to be in the ISP business you could probably find an ISP that would operate your "POP" if you provided the infrastructure with some kind of kickback arrangement.
Agreed - we aren't there yet and I doubt that we ever will be. Platter disks are still lots more expensive than RAM. Quick check of my favorites shows PC 133 SDRAM at $.07-$.9 cents per MB and big IDE drives at $.002-$.004 per MB. That is roughly a 20X-40X spread. This is an order of magnitude improvement from the 300X or so spread mid-90's when memory was $100/MB and disk was about $350/GB.
So hybrid RAM/disk solutions are desirable. But RAM does not automatically equal performance. I had an interesting conversation circa 1995 where Mike Karels (of BSD fame) talked about how you were lucky if hard drive caches didn't slow down the drive. The real performance of real file systems is a very complex problem. Subsequent to this we did some random benchmarking at the time and found about a 50/50 split where disabling the read cacheing would speed up rather than slow down real I/O performance. Part of the issue is that many drives embedded processors were not very fast so the hit of maintaining the cache was worse (since the disk to interface data path was usually a seperate fast path but cache would hit the embedded CPU).
In most OS/Filesystem combinations a file open/inital access is very expensive - lots of little I/Os to chase directories and file allocation chains until you finally get the data - then maybe a big read-ahead for you buffer cache that may or may not get used. Most filesystems are better at some things at the expense of others. So whether some amount of caching saves you any real elapsed time on an actual I/O operation is highly variable. There's also been discussion about how many filesystems (like UFS) were originally designed to try to factor in optimizations to deal with things like rotational latency and transfer speed based on rotation - that all goes out the window with any modern hard drive using zone recording/constant angular velocity (see http://www.isi.edu/netstation/zcav/zcav.html).
Right now transfer speed of the interface is a big limiting factor, as drive density goes up more and more drives can fill the pipe-at least on the outside tracks. So without new - and incompatible - interfaces you are likely to see less and less performance wins from a RAM based disk versus platter based.
IMHO - I think real wins in storage performance need to come from greater abstractions in the storage model. Imagine a device that you gave a complete path to and it then gave you a byte stream - let it manage the consistency, metadata etc. In some ways this is happening. Look at NAS versus SANs.
Network attached storage is better performance using interfaces with higher and higher bandwidths (GigE, USB 2, Firewire, Infiniband) so that it gets closer to performance with local storage.
Meanwhile local storage looks more and more like a network and has more and more features than simple block read/write (look at Storage Area Networks) - so the convergence is close in many ways.
I think real persistant storage will be using platters for a long time - until some new paradigm comes along. Look how long it's been since the venerable "Winchester" drives replaced drums, tapes, cards, etc. The RAM will be used where it's used today - as memory for CPUs to provide more intelligence and caching.
We implemented MAC (mandatory access controls) in a modified BSD/OS kernel for the Sidewinder Internet Firewall (Secure Computing) product - we adapted a scheme called "Type Enforcement" from earlier work. In essence it is just MAC capabilities and classifications with a very fine grained definition (Sidewinder ultimately had several hundred types).
In the TE - a principle (generally a unix process but occasionally a packet or device) has it's identity checked against the type of the object it is accessing. One of the privleges is the ability to change the identity and there is a state table of allowed changes - so as mentioned earlier, one key element is that most privlege changes are one way (typically from more to less). Each subsystem had it's own set of types so there was a virtual sandbox for each piece. Psuedo-objects were referenced for things like port numbers - so that a MAC could say things like telnetd can only listen on TCP port 23, but can't connect() at all or listen on other ports.
We went a step further an implemented totally seperate protocol stacks for inside versus outside so that we could give different privleges to an inside telnetd (such as allowing elevation to administrative privlege) verus and outside one.
The only form of privlege elevation was a trusted login - which had to be at the console. Even this login didn't acquire full privlege - there was a special diagnostic kernel that had full privlege (but no external networking) that could be used.
Key points we discovered over the years. One - lack of real root makes you make your software much more robust. If you have to bring down the box to get at certain privleged files you better be pretty stable. The second was how sloppy Unix software is at needing (or at least letting itself be able) to write willy nilly around the world. Without a doubt the hardest part was getting all of the userland subsystems like mail, web, login, logging, etc. to function within their mandatory little sandboxes. It was amazing how much software was sloppy - opening files that need to be read-only as writable, etc. Granted this was 7-8 years ago and things have definitely gotten better - but it still was scary.
The curious can look at U.S. patents 5,864,683 and 6,219,707. There is the usual crap in there but wade past the claims and read the description and there is good discussion as to the nature of MAC and TE and the types of attacks they help protect against.
On the plus side Unix is relatively easy to secure in this fashion. File access and network is relatively well isolated. Biggest risk is "aliasing" - different vectors that might bypass the in-kernel access checks such as mmap()ing files, inherting file handles, etc.
In general having this functionality available in open software cant help but be a good thing.
--
Mark
Slashdot Mirror
It all depends on your relationship with your manager and your trust/feel/vibe for your employer.
If you have a positive relationship and trust 3-6 months is a fairly safe window that minimizes your risks. If you are unusual you might talk about it a year or more out (I've had these discussions especially in the context of performance and development). If you are in a role where you have an individualized employment contract you might also have specific language and terms there.
If you are unsure you just wait. The challenge is firing you after discussing retirement is potentially retaliation but fighting especially at retirement time is hard. Damages are limited as your "future wages" are small. Also watch out if you are covered by binding arbitration causes - arbitrators have shameful record of siding with employers way more than the courts and even when they find for the employee providing smaller awards when they do.
Consider financial factors on timing. On the delay side like any departure also watch out for big large period (quarterly/annual) payouts - bonuses, stock vesting, 401k match, etc. If in doubt wait until your next chunk due before retirement is in the bank. What is the financial health of the company? Are there layoffs coming? Could you potentially pick up a "package" on the way out? Is there a sabbatical or other long period benefit you want to receive before you go?
On the accelerate side do you have a lot of accrued PTO/vacation to be paid out that is potentially at risk? Do you normally carry a large "float" of expenses on personal cards/$$?
Consider talking to a lawyer (and tax planner if you haven't). You may have employment agreements that cover you into retirement that may impact you especially if you plan to work on the side. Also an employment law lawyer can advise you on exactly how to bring up the topic without giving "constructive notice" (effectively the employer constitutes your inquiry as an official notice of termination of employment).
In the end - be careful and put yourself first. Many a person has had last minute retirement plans change due to unexpected shifts in financial circumstance. In a healthy organization you should be able to have this conversation well in advance - but unfortunately the window of places this type of constructive 2-way dialog is possible is shrinking. On the other hand if your skill is critical and amenable to remote and part time work you might also work out a mutually beneficial arrangement for some post-retirement income as consulting. Do you have friends or family you might refer in later (not for bonus but just for the mutual benefit of the friend and family plus company)?
In the end you will have to make a choice on risk of early notice and associated early termination against the desire to do the right thing and keep a good relationship with your former employer and co-workers.
I'm a 10+ year medic myself in a Northern rural EMS system with an average response time of 10 minutes and an average transport time of 20-25 non-emergency and 15-20 emergency. That being said we had a large coverage area and were the last ALS service on the rural fringe of a large metro area so had outlying areas as well as mutual aid calls for neighboring BLS services. So 20+ minute response times were common and my personal longest 911 (not transfer) was 52 minutes. The transport time to a definitive care facility (versus community hospital) could be an hour in good weather, 90 minutes or more in bad.
The problem with these studies is they don't do a good job of factoring transport times in the analysis and typically are conducted in high density urban areas with response times under 10 minutes. The average urban transport time is 10-15 minutes...and some urban systems approach 5-8 for ALS for a large % of their population. In rural EMS 30-45 is uncommon and sub-10 minute response times are not common.
Another problem is urban environments with Fire based non-transport ALS where theres an additional transfer of ALS care on scene to the transport crew invariably burns time. In a well coordinated system that should be minimal and the fire based EMS often provides a 3-5 minute jump on scene arrival time for EMS (typically these systems then have a somewhat lower density of transport ALS units than a pure transport only service).
Keep in mind many of the European studies are done in countries with Doctors on the rigs where they often have scene times of an hour or more...that is clearly not productive. In the US even an ALS crew on a medical should have a scene time 20 minutes unless there's scene complications. That being said for a cardiac arrest patient unless there's a very close by ER there's very little value in transporting a patient prior to resuscitation unless the scene itself has hazards and difficulties (which happens plenty).
The relevant study also focuses on cardiac arrest only and ignores the impact on ALS on various medical issues short of full arrest like breathing, allergic and diabetic emergencies as well as non-arrest cardiac emergencies. An example is several studies including one I participated in showed good ALS with pre-hospital diagnosis of a STEMI with 12-lead ECGs lead to substantially reduced onset to intervention time with the cardiac surgical team (typically angioplasty) with corresponding increase in 12 month survival rates. Another aspect is changing the protocols and cultures to bypass small community hospitals for critical patients and take them straight to hospitals with definitive care (cath lab, trauma center, etc). There's a similar argument that once a patient is in good ALS care the value add of a non-definitive care facility rarely outweighs the 1-2 hour cost of the typical admit, eval, treat, transfer cycle.
One last problem is that the 12 month survival rate for pre-hospital arrest is not great in general and has only slowly improved so focusing on small differences versus the total picture does a disservice to the value provided by the system and individual providers.
If my wife/child/mother suffers a pre-hospital cardiac arrest I am quite sure I would much rather be in a place where the timely arrival of an ALS crew is part of the response.
I've worked for a variety of companies large and small and sometimes had to take a step sideways or backwards.
Rule #1 - The only real job security you have is your skills, network and ability to find work.
Example 1: I just switched jobs 3 months ago. I worked for a stable mid-market company in a 2nd tier job West Coast market with solid profitability, above average comp for the market for the top performers, good work/life balance, and a fairly flat revenue/market cap. Everybody there acted like they were lifers - the organization got 1/2 as much done as it could have, but everybody thought it was safe and stable because they made money hand over fist on their subscription based product and did just well enough to stay in the best of breed band in their market. But the reality is every stable, profitable company has someone willing to take it and squeeze it for more. That was nine months ago, 8 months ago they announced they were being taken private, 6 months ago the deal closed, and now it's official the company HQ is moving to a mid-America 3rd tier market and it's pretty clear that the timeline for the rest of the location is end of the year at best. And since the comp is above average in an above average market with new owners that have a financial model squarely centered on keeping labor costs (well) under 50th percentile they figure nobody will take the move and the paycut so pretty much no non-executives are being even offered a relocation. Everybody who was any good and had been there less than five years have already left - the rest are left wandering the beach aimlessly waiting for the tsunami to come to shore.
So...sometimes you have to move and take a "cut" in some ways. Just remember:
Rule #2 - Nothing not in writing maters. And even that is iffy.
Got bit by that once - and if they are are small or sleazy enough even if it's in writing it doesn't matter. No point in suing when 0 chance of recovering damages and they know it.
Example 2: Took a job to "build an embedded team" (in a Java shop) with a huge pay cut. Offer letter promised stock, six and twelve month salary adjustments, and at least 3 new hires within the first six months. Company was cool, team was great, office was great. We were downtown on the river, had a pub 40 feet down the hall. But there was no money. The next "round" never closed, the stock pool was exhausted at the time the letter was written, worked a year as a team of one and left. Would have bailed sooner except the job/team/office was very cool...loved coming to work every day...but was starving figuratively and somewhat literally.
So...weigh your options but long term it pays to shift and reinvent yourself once and a while. Myself and most of my peers recognize peoples ability to reinvent themselves on a resume...as long as it's not every year :-).
Reasons to make major changes in comp and/or size:
1. Move up in title. Many (most) people who make a senior management title or exec role do it in part by taking a shift down in size or salary to move up in title and then scale up from there.
2. Validate a title or role. Maybe you want to be a manager/director/architect. Maybe every company you have had a major role at is out of business or so small nobody has heard of them. Maybe you've lead a few teams here and there but didn't have that key word in the job title. You might go into a less interesting role to validate your resume at that level and title by taking the role at a larger or more recognizable company even if work is less interesting.
3. Make more money. Maybe you've done #1 and/or #2 and now you're finally being offered that director salary and role at a big company that throws money around. You're going into your peak earning years, your 401k has been ravaged by the past decade, go for it.
4. Have fun. Maybe you've done #3, or you were lucky with your investments or you picked the right startup. You've got retirement secured, your bored, go have fun.
I'm a pilot, paramedic and software engineer. My flying is personal but I try to take a "professional" approach. I agree that there are a lot of not so great pilots out there but the most basic pilot has had a bunch more training than 99.99% of the drivers out there. A typical "commercial" pilot with a commercial certification and an instrument rating (and typically multi-engine in both) has about 3-5 times the training typically required for a CDL.
Pilots and the flying industry are one of the most regulated endeavors in modern society. A good chunk of those regulations are "written in blood" from past accidents. Besides a few thousand pages of official FAA regulations there are thousdand more in ACs, TSOs and even industry standards like SAE, Milstd, ASTM, RTCC, etc. One big part of the problem is there are no standards for UAVs or UAV operators.
Flying is still heavily dependent on "see and avoid". The reality is we probably still avoid as many or more crashes from "big sky theory" than "see and avoid". The people who want to fly UAVs mostly want to fly them where the risk is highest - down low and over population areas. Also the UAV accident rate isn't as sparse as it sounds. There are well over 100,000 hours factoring in overseas usage. Even if you subtract out combat or unknown losses the accident rate of the UAV business is abysmal. Remember this is the industry that gave us unencrypted classified combat video. Check out http://www.homeland1.com/homeland-security-products/unmanned-aerial-vehicles-uav/articles/847069-accident-reports-show-us-drone-aircraft-plagued-with-problems/.
The argument that there is no pilot so the risk is minimal is disturbing. A predator is almost 30' long and a 48' wingspan - 1200# empty and over a ton fully loaded. This is comparable to most 4 seat trainers. Several of the private drones are smaller but have even less QA and little to no redundancy.
The first and only NSTB report on a drone crash is at http://www.ntsb.gov/ntsb/brief2.asp?ev_id=20060509X00531&ntsbno=CHI06MA121&akey=1. It cites the typical chain of errors as well as a series of poor design decisions. It also notes the wreckage path indicated a flat approach and a wreckage path of almost 100' with jet fuel scattered around the crash site (there was no ignition). The operations console being used had suffered approximately 16 lockups in the 4 months prior to the crash and suck lockups were viewed as normal and acceptable. The normal "lost link" procedure normally keeps the aircraft flying a predetermined route over unpopulated areas until control was restablished but improper recovery on the crash failure caused the engine to be turned off.
The lost link route procedure was called out in the NTSB report: "Another contractor, Organizational Strategies, Inc. (OSI), provided the coordinates for the lost-link waypoints to CBP. OSI reported that it developed the waypoints using an Internet satellite website. CBP reported that it also used the same Internet satellite website to verify the location of the waypoints. According to this website, some of the website's imagery is 1 to 3 years old. Neither OSI nor CBP used additional methods to confirm that the waypoints were not located over populated areas." No indication of the resolution of the satellite imagry used - and no requirement for direct verification.
In fairness the CBP is actually one of the more rigourous operators of UAVs. Their pilots are required to be certificated pilots with at least 200 hours of actualy flying time and 200 hours of UAS flying time. They also use specific TFRs to provide seperation and maintain contact and obtain clearances from ATC. Not all FAA "Certificates of Authority" require this level of coordination or training. Many smaller operators operate close enough to the ground or restricted terrain or existing restricted airspace viewed to not interfere with existing flight activity.
The simple reality is the UAV industry is about where manned flight was in the 30s. They hav
I'm going to avoid the topic of the relative goodness/badness of patents and address the question at hand.
Bonuses in my experience have been in the range of $1k-$2k for filing and $2k-$5k for granted. As noted above a big issues is what is the policy for patents granted to former employee inventors. I have missed out on all of my granted bonuses because all of my patents have been granted after I had moved on.
I would definitely have a cash component to the bounus - most of us have gotten savvy enough about the real odds on options payoff so it isn't necessarily a reward. At this point I view options as a quid pro quo for risk but not a substitute for fair salary and at least semi-reasonable workload.
The keeping in touch factor is a real issue - successful patents often involve one or more refilings and access to the inventor can be very helpful in adjusting claims or clarifying details. One former employer had the typical policy of not paying bounus for employees after they left. They had gotten back in touch with me about taking time to review a redone application which would have involved several hours of work. Given the lack of any reward compenent I requested a basic consulting rate for my time and they refused to I pretty much told them they were on their own.
A more recent former employer has a policy of paying bonuses to former employees if the employee can be reached within two years of the bonus event and providing they were cooperative in assisting any follow up reviews. Given that I have been cooperative and helpful with a couple of review and clarifications along the way.
Last but not least some places give a bonus premium on the "primary" versus additional inventors.
So in developing your policy pick amounts and make sure you are explicit either way about the bonus disposition for departed employees. The bonus progam can be an incentive for the innovators in the company as well as a tool to help get cooperation on the followup even if they have moved on.
--
Mark
I'm in the same boat. I run a small shop with 10 folks and Exchange seems like overkill. Every year or so I go kicking around for an alternative and come back to staying put.
I'm not an MS bigot - there's FreeBSD and Linux (currently CentOS) doing the bulk of the server work here outside of groupware, but the key features in Exchange I have a hard time finding elsewhere is:
-The strong calendaring/scheduling including resource management
-The good quality web access (admittedly OWA requires IE to really shine)
-Support for mobile devices (between OMA/ActiveSync and IMAP I can almost always make it work)
-Easy to manage (at my small scale) replication
-Notes and Tasklists functionality does get used a lot in our organization
-Relatively seemless support for online/offline via Outlook
-Public folders w/support for offline and the various data items
-Rich, flexible and easy to manage ACLs on public and private folders
-We do a lot with shared calendars, notes and task lists
Various bits out there have pieces of the above but the real thing I can't seem to find is the strong calender/scheduling features with offline support. The reality is for serious road warriors the internet isn't anywhere close to 100% available. Remote areas, client sites with restrictive network access policies time in the airplane, all make 100% online solutions unworkable for us so far.
But as a small organization the only way we do this and stay legal is as a Microsoft partner through our NFR product licenses, the fact that we are 10 people makes it work. We couldn't afford the $1000s a year it would cost us to stay legal through retail or the lower volume license plans.
Like the rest of the world Exchange 2007 is on the shelf for now even though we have access to licenses for it. One is the 64 bit requirement along with overall performance loads raises the bar to deplying it. Microsoft is doing a typical MS thing and moving away from a good solution because of limitations before the next thing is ready. The de-emphasis of Public Folders and the threats to drop them completely are discouraging. Their answer is "SharePoint" which has no replication and no offline abilities currently - plus it forces me to a different tool and technology when all I need is to share access to various messaging items.
In practice I think MS is effectively priced itself out "small medium" businesses. The only way to deploy the Exchange, the required Windows server, CALs, etc. affordably is through a partner plan like us (which only scales to 10 users) or through Small Business Server - which doesn't scale effectively beyond 25-50 users and imposes deployment and licensing limitations that make it very difficult to move to the next level once you hit that ceiling.
Last but not least to that switch is data migration. I can kludge around on the messaging with various IMAP based solutions to moving the messages and but expect I'll have to write off the years of useful history in calendars. Also contact migration with LDIF still never seems to quite live up to promise, things always get munged, stripped or loose some supporting data.
There's plenty of slings and arrows at people who equate "Open" with "Free", and I agree you don't get something for nothing. Bur for many us it's a basic issue of affordability combined with legality. We work very hard to stay legal but even with our small shop licensing would eat us alive without the available partner program options and using open source in some areas. Everyone wants to grow their revenue and things creep up over time. Anti-spam, Anti-virus, office suites, server licenses, CALS, etc. Last time we ran numbers it would cost us over $2000/yr per/user to stay legal with retail licenses if we fully drank the MS cool aid. MS is getting pressure to prune their partner lists fr
Ooops, yeah, my bad, MM's correction is correct. But as AC notes, they have MySQL AB in a spot, especially now that they _also_ have bought SleepyCat/Berkeley DB.
If you read the attached article they're really talking about the performance deficit of PHP on Windows. Running PHP on Windows you have two primary server options - IIS and Apache, and then the various integration strategies (plug-in, CGI, FastCGI). In the case of IIS plug-ins with ISAPI there's a number of fiddling bits (like the ISAPI extension versus filter stuff) and getting all the settings right and quirky bits (example - I've never been able to mix PHP4 and PHP5 on one server with ISAPI). The most common setup is FastCGI - traditional CGI on Windows is painful because Windows process creation overhead is a lot more expensive - but FastCGI imposes tradeoffs on resources and performance and numbers of servers, etc.
.NET down your throat.
All in all as a frequent user of PHP I figure this will benefit most of us. I worry a little less about extend and coopt here because I think Microsoft's primary motivation is to get you to stay on Windows Server/IIS even for your PHP platforms and then continue to try to push
As for MySQL - it's now owned by Oracle and IMHO Larry Ellison has a far better shot at being the antichrist than Bill Gates. Yes we have all that GPLd code but the company, talent and non-GPL rights to the code are owned by Oracle.
Another aside is both cases are examples that successful technology attracts attention and money regardless of the ideology or goals that spawned it. As someone who's been plugging away in development almost 20 years now I'd say the "success" part is tied more to market success and less to technological superiority - MySQL and PHP both end up strong on both counts, but neither would have attracted a whit's bit of interest if their market penetration was 2%.
I know this reply is way late in the arc on this post but the threads here hit home so very, very much I figured I'd post to an empty room anyway.
I've been on both sides of the fence as top talent and trying to hire top talent. First the hiring view. The reality is in certain niches (especially networking, protocol, embedded systems type talent) there aren't nearly as many rock stars as you think. My last permanent job was smack in the middle of the Valley and the reality is there were probably only a couple of hundred people out there in top little piece of the bell curve in any general skill set. There's a ton of talented people just a bit farther down the curve, but you look at most tech companies and it's really a very, very small group of people that drive the design, technology and innovation. It also seemed like you had to find the rare combination of technically brillant but market aware individual as the quality of product marketing in most organizations is _sh*t_ as far as defining innovation (versus spitting out glossies extolling your virtues and trashing the competition).
Then you can't hire them because you can't really get management to pay the top salaries because the VCs are breathing down the necks of the officers and board wondering why it hasn't already all been shipped to India much less trying to shake out an attractive salary.
On the working side once you get hired you end up only being around 2-4 years on average because you get squeezed on the late side of the capital/startup curve and salaries stagnate at 2-3% a year and bonuses and perks get squeezed - or you just plain run out of $$ and go under and if you're good enough you eventually become the most expensive body in the shop and easy pickings in a quarter the beancounters are sweating dollars instead of head count.
I've worked for a ton of startups as well as established companies and the reality is 98% of companies more than 10-20 people as an organization don't give a rat's a** about their people. There's still individuals out there that fight the good fight but the reality is if they need to do a short pump and dump on the share price or fixup the burn burn rate prior to the next financing round they'll RIF us 20% accross the board in a heartbeat and then whine a month later about the lack of employee loyalty these days and wonder why we jump ship for a lousy 5% salary differential or an extra week of vacation. Options are worthless - I've had some come in and most go under and my net gain from options over a 20 year career works out to about $1500 a year (maybe $2000 in today's dollars).
There's a glass ceiling of $100-$120k a year everywhere outside the Valley and maybe $140k a year in the Valley (where the delta doesn't cover the cost of living diffs) if you want to be a developer, architect, technical anything. And that's the very peak of the bell curve. There's a bunch of development jobs that have typical _Senior_ developer salaries down more in the $60-$80k and the non-developer IT guys have it worse. To move forward you have to cross the dark side to management or leave the traditional work force entirely and do your own thing - consulting, starting a business, inventing the next pet rock. Out here in middle America the number development focused technical companies is small enough that there aren't a lot of Director and up level positions - and the majority of them are in larger organizations that tend to promote from within so to break that inital barrier into management requires backsliding your salary - sometimes a lot - and dredging along at "Director of IT" for a salary that works out to a pitiful amount per hour just to slide the resume over to a new track with job titles with words like Director and Vice in them.
There's a lot of top flight technical talent outside the Valley and the Beltway that will choose to schluck along under-employed just to live in a place with affordable housing and decent public schools and a bit of green grass, but most companies are so weakl
I used to live in Pleasanton, California and would fly out of Livermore into the former Castle Airforce Base in Atwater, CA which was home to 93rd Bombardment Wing's B-52s for almost 30 years (Castle Airport). It's now officially 11,802 by 150, in reality the runway is 13,000 by 300 but they striped it down to reduce maintenance costs. It's an uncontrolled field but it does get some use for flight testing and training and the FBO there does do heavy maintenance so I have shared the pattern with a 737 once and a DC-10 another time. The big empty B-52 hangers are open to walk through - it's interesting looking at the "wash stand" for B-52s. When the Air Force was still there (up through 1997) they used to host the Golden West EAA Fly-Ins - for that they would split the runway in two and use each end at the same time. There's a few 100 acres of paved tarmac and they do routinely close off part of it for road rallys and other driving schools. There's a nice museum on field (Castle Air Museum) and the FBO will shuttle you there for free, the walk is about 1/2 mile. The museum itself runs a few bucks but is worth it.
I work part time as a paramedic. I agree with most posters, after 50-60 hours a week of startup grind the last thing I want to do is IT or programming for someone else. What little spare energy I have goes into my personal bits for myself.
Being a paramedic is everything IT is not - lot's of people interaction, some physical labor, outside time. It's also personally rewarding in a way that programming isn't. I've never been applauded for my work as a programmer, something I have had happen a few times as a medic. I've also never been offered a blow job as a programmer, but I have been a few times as a medic. As a married and ethical man I do not require applause nor accept the blow jobs but as a human being and a terminal male I appreciate the consideration present in both.
I originally became a medic because a bit of volunteering as an EMT showed me I loved the work and I thought that (as my plans were at the time) that the medic job would provide a salary baseline and benefits when consulting wore thin. Well - the bubble burst and I'm an employee again but I keep up the medic because it's a perfect escape from the office and if things really go south it's one job I know they _can't_ ship to India.
The wages are not great but when you can work a 24 hour shift and get paid 24 hours for one calendar day it does add up even for a few days a month enough for even an overpaid technoweenie like me to notice.
After 15 years in the development trenches I would love to work full time as a medic and have the energy to expend my skills part time on programming but then I couldn't afford a new GeForce 6600 or flying as often. I also like not sweating that I'm getting the $4 latte instead of the $1.25 cup of joe, something most full time medics have to worry about.
At some point I may make the trade of money versus time for lifestyle as we achieve certain financial goals, but for now it works as is.
I'm a private pilot and have several friends and relatives who work as pilots from the lowly cargo dogs flying single engine FedEx to a DC-10 captain at American Airlines.
/. crowd should know more than most how hard it is to program reponses for when the situation departs "normal" parameters.
One - the $250k number probably represents the top 5% of pilots for compensation. Even the pilots earning those numbers took 10-15 years and spent a whole lot of $$ to get there - makes Med School look fast and cheap.
Two - many (not all) pilots are paid all or mostly by the hour flown - they aren't paid for the hours of flight planning and prep or the days away from home. An international flight crew might easily be away from home 20 out of 30 days and that is gone for 24 hours, no family, maybe one phone call a day if the time you are on the ground lines up with the time your family is awake. Out of that remaining 10 days they may still have to get in their periodic training, reviews, etc.
Three - yes computers can land the airplane all the way to the pavement. But they can only do it (today) at a handful of runways at a handful of airports with the right precision (Cat III ILS) approach. As a point of reference in all of California - the home of the most airports overall and the most airports which scheduled air service of any of the US states there are less than a dozen Cat III ILS approaches. Even that number is unusually large because of the relatively small (towns like Fresno) cities with Cat III because of the unique weather of the Central Valley (Tule Fog).
Over the other 98% of the world you need a pilot to keep flying and land. Technology is rapidly improving and things like GPS will make more and more precision approaches possible but it will still take a pilot to land at the majority of runway ends in the world. And when shit goes wrong you want to land at the closest runway that you can get safely stopped in - not the one that the computer can use 30 minutes or an hour or more farther away.
Computers are great at doing things requiring fast calculations and precise responses but the
Granted - people don't always do better - they typically still need some training and guidance for how to respond and tend to do poorly when flight departs completely from the norm. Look at the end result of events like the accidental thrust reverser deployments in 767s or the rudder failures in 737s. But in many of these events even the post-mortem reviews shows that the "window" to recover from these events is often 10 seconds or less. But other situations like the Souix City crash or the volcanic ash incident of Speedbird 1 (British Airways) show that the adaptability of a human crew is irreplacable.
It is probably inevitable that computers will do more and more of the flying - they already do 80%-90% of the enroute flying and they typically fly most of the hard appraochees (at least on the "big" guys) with the pilot acting more as a director and flight computer programmer than a pilot. But think - as the pilots really fly less and less then how do they maintain their skills for the 5% of the time that they are really needed?
So I think the pilots of the world earn every nickel the hard way.
I tend to agree...
In past lives I've worked with career spooks and even had a few of my own visits to the NSA (no I've never been to Fort Meade - any real spook groupie knows that the Infosec guys in "X" were up at the "Friendship Annex" by BWI...).
But it was an infrequent but not common practice at vendors that some systems to certain customers would be delivered by leaving a trailer at location X where it would disappear for a few days then they would get a call to pick up the trailer at Y.
There is also a verification process to sell to private and foreign customers to verify that systems aren't going to the wrong people.
But the real question is that now that Cray is Cray again will they start shipping a case of Leine's with every system like they used to? (Probably not given that manufacturing is no longer in Chippewa Falls, WI).
For travesties in interstate naming in the Bay Area I-980 is just as bad - would it have been that bad to call it CA-24 for another mile and a half.
I'll also be the first to agree that in the cheap PC hardware era many performance problems can be most effectively solved by throwing hardware at them. An example would be a CPU bound application that needs to be twice as fast that runs on 1 PC server with 9 month old technology could easily make that target with a $3000-$5000 hardware upgrade. That doesn't by very many man hours to be sure.
I wasn't (and wouldn't) claim that a RAM disk wouldn't be faster than a platter based drive - only that the amount of that gain is a complex question. RAM drives for certain point applications (lock dirs, temporary files, caches, etc) are great. Even Platypus which you mention makes a RAM drive product that doesn't have persitent storage (and admittedly some that do).
So just a quick followup. I wouldn't have bothered except that it's my first post to be replied to and my first post to be moderated. Woo hoo.
There are a number of possible solutions.
.
Wireless is one - there are a number of 10BaseT to DS3 (ie: 45Mbps) products available on the Market. Link price runs $30,000-$60,000 (link=2X antennas+boxes, etc). These solutions generally top out at 10-20 miles, a few do 30. So you could do it in 2-3 hops. Look for products that operate in unlicensed spectrum (typically 2.4GHz and 5.8 GHz Uniband) to speed deployment. Harris Microwave (http://www.microwave.harris.com/) and Western Multiplex (http://www.wmux.com/) are two that come to mind.
Another option is free space optical (fso) - companies like Lightwave (www.lightwave.com).
Higher bandwidth wireless and require line of sight. RF in fact requires more clearance - based on a function called the Fresnel zone radius - a shape resembling a stretched football - that makes required altitudes for longer links even higher. General rule is that 60% of the 1st fresnel zone must be unobstructed. At 45 miles and 2.4 GHz the middle of the zone typically requires a clearance of 100'-160'. At 5.8GHz this drops to 60'-100' roughly (the function is proportional to wavelength and thus inversely proportional to frequency) - one calculator is at http://wireless.ydi.com/calc.asp#FresnelZone - another simpler one is at http://www.ingenious-nets.com/fresnel-calc.html.
But a number of people claim getting 20 mile ranges out of 802.11b at 2Mbps rates. One sample calculation gives the required antenna height at abou 175 feet (about 100 feet for line of sight and 75 feet for the fresnel zone clearance) (see http://www.oreillynet.com/cs/weblog/view/wlg/197)
A 175' antenna height is potentially very expensive. More expensive than typical antennas for normal RF applications because of the fact that the antenna itself must be very solid in wind to maintain line of sight alignment. But a some water towers or a 15 story building would do it.
Net is that for RF wireless you probably need to drop to more intermediate links to manage LOS and clearance.
But there may be viable wireline alternatives. One source of help would be a 'national' ISP like MCI/Worldcom (UUNet), Sprint/Earthlink, or Verizon (GTE) (in the case of the US). Their local sales forces are often very in touch with various alternative data carriers to provide the backhaul. Pipeline companies, dedicated fibre companies and power companies are among the entities that often have a datalink business.
Almost any local carrier can provide at least a T1 level frame relay or ATM circuit. One nice technology for this application is Inverse-Multiplexed ATM. For instance - you take 3 T1 ATM circuits and run IMA over them and the link looks like one ATM link at 4.5Mbps. If individual links go up or down the IMA link stays up - the bandwidth just goes up or down.
There are federal programs available to help with school internet access. Many states also have programs geared towards both education and general rural data access.
The advice about a Co-op is good. If you don't want to be in the ISP business you could probably find an ISP that would operate your "POP" if you provided the infrastructure with some kind of kickback arrangement.
Good luck.
Agreed - we aren't there yet and I doubt that we ever will be. Platter disks are still lots more expensive than RAM. Quick check of my favorites shows PC 133 SDRAM at $.07-$.9 cents per MB and big IDE drives at $.002-$.004 per MB. That is roughly a 20X-40X spread. This is an order of magnitude improvement from the 300X or so spread mid-90's when memory was $100/MB and disk was about $350/GB.
So hybrid RAM/disk solutions are desirable. But RAM does not automatically equal performance. I had an interesting conversation circa 1995 where Mike Karels (of BSD fame) talked about how you were lucky if hard drive caches didn't slow down the drive. The real performance of real file systems is a very complex problem. Subsequent to this we did some random benchmarking at the time and found about a 50/50 split where disabling the read cacheing would speed up rather than slow down real I/O performance. Part of the issue is that many drives embedded processors were not very fast so the hit of maintaining the cache was worse (since the disk to interface data path was usually a seperate fast path but cache would hit the embedded CPU).
In most OS/Filesystem combinations a file open/inital access is very expensive - lots of little I/Os to chase directories and file allocation chains until you finally get the data - then maybe a big read-ahead for you buffer cache that may or may not get used. Most filesystems are better at some things at the expense of others. So whether some amount of caching saves you any real elapsed time on an actual I/O operation is highly variable. There's also been discussion about how many filesystems (like UFS) were originally designed to try to factor in optimizations to deal with things like rotational latency and transfer speed based on rotation - that all goes out the window with any modern hard drive using zone recording/constant angular velocity (see http://www.isi.edu/netstation/zcav/zcav.html).
Right now transfer speed of the interface is a big limiting factor, as drive density goes up more and more drives can fill the pipe-at least on the outside tracks. So without new - and incompatible - interfaces you are likely to see less and less performance wins from a RAM based disk versus platter based.
IMHO - I think real wins in storage performance need to come from greater abstractions in the storage model. Imagine a device that you gave a complete path to and it then gave you a byte stream - let it manage the consistency, metadata etc. In some ways this is happening. Look at NAS versus SANs.
Network attached storage is better performance using interfaces with higher and higher bandwidths (GigE, USB 2, Firewire, Infiniband) so that it gets closer to performance with local storage.
Meanwhile local storage looks more and more like a network and has more and more features than simple block read/write (look at Storage Area Networks) - so the convergence is close in many ways.
I think real persistant storage will be using platters for a long time - until some new paradigm comes along. Look how long it's been since the venerable "Winchester" drives replaced drums, tapes, cards, etc. The RAM will be used where it's used today - as memory for CPUs to provide more intelligence and caching.
We implemented MAC (mandatory access controls) in a modified BSD/OS kernel for the Sidewinder Internet Firewall (Secure Computing) product - we adapted a scheme called "Type Enforcement" from earlier work. In essence it is just MAC capabilities and classifications with a very fine grained definition (Sidewinder ultimately had several hundred types).
In the TE - a principle (generally a unix process but occasionally a packet or device) has it's identity checked against the type of the object it is accessing. One of the privleges is the ability to change the identity and there is a state table of allowed changes - so as mentioned earlier, one key element is that most privlege changes are one way (typically from more to less). Each subsystem had it's own set of types so there was a virtual sandbox for each piece. Psuedo-objects were referenced for things like port numbers - so that a MAC could say things like telnetd can only listen on TCP port 23, but can't connect() at all or listen on other ports.
We went a step further an implemented totally seperate protocol stacks for inside versus outside so that we could give different privleges to an inside telnetd (such as allowing elevation to administrative privlege) verus and outside one.
The only form of privlege elevation was a trusted login - which had to be at the console. Even this login didn't acquire full privlege - there was a special diagnostic kernel that had full privlege (but no external networking) that could be used.
Key points we discovered over the years. One - lack of real root makes you make your software much more robust. If you have to bring down the box to get at certain privleged files you better be pretty stable. The second was how sloppy Unix software is at needing (or at least letting itself be able) to write willy nilly around the world. Without a doubt the hardest part was getting all of the userland subsystems like mail, web, login, logging, etc. to function within their mandatory little sandboxes. It was amazing how much software was sloppy - opening files that need to be read-only as writable, etc. Granted this was 7-8 years ago and things have definitely gotten better - but it still was scary.
The curious can look at U.S. patents 5,864,683 and 6,219,707. There is the usual crap in there but wade past the claims and read the description and there is good discussion as to the nature of MAC and TE and the types of attacks they help protect against.
On the plus side Unix is relatively easy to secure in this fashion. File access and network is relatively well isolated. Biggest risk is "aliasing" - different vectors that might bypass the in-kernel access checks such as mmap()ing files, inherting file handles, etc.
In general having this functionality available in open software cant help but be a good thing.
--
Mark